150 Lahy and Roberts: 



placed in the rotating field. The axes A and B are brought into 

 coincidence. The determination of J then becomes a question of 

 finding the work done on the copper cylinder, and the lieat deve- 

 loped in it. The work is found as in Rowland's experiment. The 

 copper cylinder is attached to a vertical axle and wheel (diameter, 

 D cm); and the whole is suspended by a torsion wire. I'wo wires 

 pass round the circumference of the wheel over pulleys to two 

 masses of m gm. The couple mgD dyne. cm. produced by these 

 weights balances the couple arising from the rotating magnetic 

 field. The torsion wire gives stability to the system. 



The heat developed is measured by a continuous flow method. 

 Water' flows past a platinum thermometer, then it circulates round 

 the copper cylinder and out paist a second platinum thermometer. 

 The heat developed is M(6'2 — ^i) calorie, where M gm. is the 

 quantity of water flowing in an experiment, and (^2 — ^1) degree 

 is the rise of temperature. To eliminate the heat losses, L calorie, 

 from the expression- for J, two experiments are made — a heavy and 

 a light one — in which the inlet and outlet temperatures' Oy and 6^ 

 are the same. The heat developed in, the former is about ten times 

 that in the latter. We then have for the heavy experiment 



7rD7/mi(7 = J(^,-^i)Mi + L 

 and for the light 



7rDn?«2^ = J(^,-^i)M,-|-L 

 and therefore 



J =7rDn^(m, - .;,,)/ \ (M, - ^\,){e, - 0,) } 



where n is the number of revolutions of the field magnets for the 

 period of an experiment, which is the same for the light and heavy 

 experiments. 



The heat developed in the light experiment is made less than in 

 the heavy experiment by reducing the strength of the magnetic 

 field. 



In the above relation it is assumed that the loss of heat, L, is 

 constant for a given value of the inlet and outlet water tempera- 

 tures. The text-book accounts of Callendar and Barnes' experiments* 

 lay inadequate stress on the conditions which those observers 

 showed must be fulfilled for this assumption to be justified. In our 

 preliminary experiments the loss of heat, L, bore little relation to 

 the inlet and outlet water temperatures, Oy and 6.>. With $y and 

 ^2 fixed, L would vary widely with the rate of flow of the water 

 through the copper, which was then in the form of a hollow copper 

 ring, xifter the factors which determined the heat losses in this 

 form of the apparatus had been determined by a number of experi- 



3. Callendar, Phil. Trans.. Vol. CXCIX., 1902, pp. 112, 114. 115, 122,. 

 Bcirnes. Phil. Trans., Vol. CXCIX., 1902. pp. 224-228. 



